Tea consumption and risk of cancer of the colon and rectum.

Abstract

The association between tea consumption and risk of colon and rectal cancers was investigated in a population-based case-control study conducted in Iowa (United States). Colon (n = 685) and rectal (n = 655) cancer cases age 40-85 yr were identified through the Iowa Surveillance, Epidemiology, and End Results (SEER) Cancer Registry (86% response rate); controls (n = 2,434) were frequency matched by sex and 5-yr age group (80% response rate). The usual adult consumption of tea (hot and iced), along with other information including dietary data, was self-reported using a mailed questionnaire. Total tea consumption (cups/day) was categorized as none (reference category), low (< 3.1), medium (3.1-5.0), and high (> 5.0), with cut points for tea consumers based on the 75th and 90th percentiles of use among controls. Unconditional logistic regression was used to estimate the odds ratios (ORs) and 95% confidence intervals. There was no association between total tea consumption and colon cancer (ORs = 1.0, 1.1, 1.3, and 0.7) or rectal cancer (ORs = 1.0, 0.9, 1.4, and 1.0) after adjustment for age, sex, education, physical activity, smoking history, and intake of coffee, fiber, and fruits and vegetables. Results were similar when hot tea and iced tea were evaluated individually. Further adjustment for other colorectal cancer risk factors did not alter these results. There was no association with proximal or distal colon cancer. There was also no interaction between tea consumption and any of the dietary variables or total fluid on risk of colon or rectal cancer, with the exception of a suggestive positive association between an increasing frequency of tea consumption and colon cancer risk among current smokers (multivariate ORs = 1.0, 1.4, 2.0, and 1.8; P for trend = 0.1), but not among never smokers (multivariate ORs = 1.0, 1.0, 1.1, and 0.4; P for trend = 0.3). These data do not support an overall association, either positive or negative, between tea consumption and risk of colon or rectal cancer in this Mid-western US population.

Full text

Page 1

Tea Consumption and Risk of Cancer of the Colon and Rectum
James R. Cerhan, Shannon D. Putnam, Greg D. Bianchi,
Alexander S. Parker, Charles F. Lynch, and Kenneth P. Cantor
Abstract: The association between tea consumption and risk
of colon and rectal cancers was investigated in a popula-
tion-based case-control study conducted in Iowa (United
States). Colon (n = 685) and rectal (n = 655) cancer cases
age 40–85 yr were identified through the Iowa Surveillance,
Epidemiology, and End Results (SEER) Cancer Registry
(86% response rate); controls (n = 2,434) were frequency
matched by sex and 5-yr age group (80% response rate).
The usual adult consumption of tea (hot and iced), along
with other information including dietary data, was self-
reported using a mailed questionnaire. Total tea consump-
tion (cups/day) was categorized as none (reference cate-
gory), low (?3.1), medium (3.1–5.0), and high (?5.0), with
cut points for tea consumers based on the 75th and 90th per-
centiles of use among controls. Unconditional logistic re-
gression was used to estimate the odds ratios (ORs) and
95% confidence intervals. There was no association be-
tween total tea consumption and colon cancer (ORs = 1.0,
1.1, 1.3, and 0.7) or rectal cancer (ORs = 1.0, 0.9, 1.4, and
1.0) after adjustment for age, sex, education, physical activ-
ity, smoking history, and intake of coffee, fiber, and fruits
and vegetables. Results were similar when hot tea and iced
tea were evaluated individually. Further adjustment for
other colorectal cancer risk factors did not alter these re-
sults. There was no association with proximal or distal colon
cancer. There was also no interaction between tea consump-
tion and any of the dietary variables or total fluid on risk of
colon or rectal cancer, with the exception of a suggestive
positive association between an increasing frequency of tea
consumption and colon cancer risk among current smokers
(multivariate ORs = 1.0, 1.4, 2.0, and 1.8; P for trend = 0.1),
but not among never smokers (multivariate ORs = 1.0, 1.0,
1.1, and 0.4; P for trend = 0.3). These data do not support an
overall association, either positive or negative, between tea
consumption and risk of colon or rectal cancer in this Mid-
western US population.
Introduction
Tea (camellia sinensis) is one of the most commonly con-
sumed beverages in the world. Tea contains several poly-
phenolic compounds with potent antioxidant properties that
exert a protective effect in a number of experimental tumor
models, including colorectal cancer (1–3). Whereas most
experimental studies used green tea extracts, black tea, the
form most commonly consumed in Western countries, also
appears to show chemopreventive activities for a variety of
tumor models (3–5). In contrast, epidemiological data are
more limited and, to date, are equivocal regarding a protec-
tive association of tea consumption, black or green, and can-
cer risk in general (1,2).
Epidemiological studies have generally found no consis-
tent association between risk of colorectal cancer and black
tea consumption (6–20). However, of ?15 studies evaluating
the association of black tea and colorectal cancer, most have
used relatively weak study designs, including ecological
(difficulty extrapolating group data to individuals), hospital-
based case-control (controls likely to be biased in tea
consumption), or cohort studies with mortality end points
(inability to distinguish effects on cancer incidence vs. sur-
vival). After these studies are excluded, there are only two
population-based case-control studies and four cohort stud-
ies of colon or rectal cancer incidence, only two of which
were conducted in the United States (Iowa and Hawaii), sup-
porting the need for additional studies using stronger study
designs to evaluate this association. Furthermore, most prior
studies have not distinguished hot tea from iced tea con-
sumption, and polyphenol content can vary between these
beverages (21,22).
We evaluated consumption of hot and iced teas (assumed
to be predominantly black tea) and risk of colon and rectal
cancers in a population-based, case-control study. We were
able to control for potential confounding by other colorectal
NUTRITION AND CANCER, 41(1&2), 33–40
Copyright © 2001, Lawrence Erlbaum Associates, Inc.
J. R. Cerhan and A. S. Parker are affiliated with the Department of Health Sciences Research, Mayo Clinic, Rochester, MN 55905. S. D. Putnam and C. F.
Lynch are affiliated with the Department of Epidemiology, University of Iowa College of Public Health, Iowa City, IA 52242. G. D. Bianchi is affiliated with
the Department of Urology, University of Iowa Hospitals and Clinics, Iowa City, IA 52242. K. P. Cantor is affiliated with the Occupational Epidemiology
Branch, National Cancer Institute, Bethesda, MD 20892.

Page 2

cancer risk factors and evaluate potential interactions that
have been previously reported or are biologically plausible,
including those with age, smoking, dietary factors, and total
fluid intake (10,14,19).
Subjects and Methods
Sample Population
This population-based case-control study was conducted
between 1986 and 1989 with a major focus on the associa-
tion of drinking water contaminants and cancer risk at six
cancer sites. Further details of the study with regard to colon
and rectal cancers are published elsewhere (23), as are the
results for tea consumption and risk of bladder and kidney
cancer (24). Briefly, incident cases of colon cancer (diag-
nosed March–December 1987) and rectal cancer (diagnosed
January 1986–December 1987) were identified through the
Iowa Cancer Registry, which is part of the National Cancer
Institute’s Surveillance, Epidemiology, and End Results
(SEER) program (25). After physician consent was obtained
to contact cases, a letter was sent to each case or next of kin
explaining the study. Eligibility criteria included Iowa resi-
dency at diagnosis, age at diagnosis between 40 and 85 yr,
histological confirmation of the cancer, and no previous his-
tory of malignant neoplasms, excluding nonmelanoma skin
cancers. The participation rate for the 685 colon cancer cases
was 86%; 95 (14%) were proxy respondents. The participa-
tion rate for the 655 rectal cancer cases was also 86%; 115
(18%) were proxy respondents.
A single control series was used in the analysis for both
cancer sites. Cases from all six cancer sites were frequency
matched to controls by sex and 5-yr age group, with a case-
control matching ratio of ~2.3:1. Controls ?65 yr of age
were selected from the state of Iowa driver’s license records,
whereas those aged 65–85 yr were selected from rosters of
the US Health Care Financing Administration. Both sources
have very high coverage of the Iowa population (26,27). The
participation rate for the 2,434 controls was 80%; 2 were
proxy respondents.
Data Collection
Eligible cases and controls (or their proxies) were con-
tacted by telephone and invited to participate by completing
a self-administered, mailed questionnaire. The question-
naire ascertained demographic data, occupational history,
smoking history, familial history of cancer in first-degree
relatives, and past medical conditions. A 55-item food fre-
quency questionnaire (usual adult consumption) was also
included, as well as detailed items regarding adult fluid con-
sumption. Specific questions were asked about the usual
adult consumption of iced tea and hot tea (cups/day) as well
as coffee and other beverages at home and outside the home.
Respondents were specifically asked to report what they had
consumed over all their adult years and exclude any changes
in the last couple of years. Total beverage consumption was
estimated from usual adult fluid intake of drinking water,
coffee, tea, fruit juices/drinks, soups, milk, soft drinks, and
alcoholic beverages, as previously reported (24).
Subjects reluctant to complete the detailed questionnaire
(7.7%) were offered a 15-min abbreviated telephone inter-
view that included information on beverage (including tea)
consumption, smoking, and other critical information for the
analysis of water quality and cancer risk but excluded the de-
tailed occupational history and food frequency question-
naire.
Statistical Analysis
Of the 3,774 participants who participated in the study,
we excluded 102 (2.7%) who were missing data for tea con-
sumption. Hot tea consumption was categorized as none,
low (?1.9 cups/day), medium (1.9–3.0 cups/day), and high
(?3 cups/day), with cut points for users based on the 75th
and 90th percentiles of consumption among controls. Using
the same percentiles, iced tea was categorized into no, low
(?3.0 cups/day), medium (3–5 cups/day), and high (?5 cups/
day) consumption. Because most previous studies did not
distinguish hot tea and iced tea, we also created a total tea
variable, which was the sum for consumption of hot and iced
teas. This variable was categorized into no, low (?3.1 cups/
day), medium (3.1–5.0 cups/day) and high (?5.0 cups/day)
consumption, again based on the 75th and 90th percentiles
of consumption among the controls.
Unconditional logistic regression was used to estimate
the odds ratios (ORs) and corresponding 95% confidence in-
tervals (CIs) for the association between tea consumption
and risk of colon and rectal cancers. Colon cancer was fur-
ther subdivided into proximal cancer (cancers of the cecum,
ascending colon, hepatic flexure, and transverse colon) and
distal cancer (cancers of the splenic flexure, descending co-
lon, and sigmoid colon) and was based on the SEER ICD-O
coding of the cancer site (28). Eighteen colon cancer cases
were excluded from the subsite analysis, because their sub-
site was “appendix,” “colon, not otherwise specified,” or
“overlapping lesions of the colon.”
ORs were initially adjusted for age and sex. We also eval-
uated sex-specific models, but there was little heterogeneity
in results (data not shown), so only models with both sexes
combined are presented. All ORs were then adjusted for
confounders in this data set (i.e., factors associated with tea
consumption and colorectal cancer risk). Finally, models
were fitted that included an extended number of other fac-
tors associated with tea consumption or risk factors for colo-
rectal cancer.
We also evaluated the possible modifying effects of age
(?69 and ?69 yr), first-degree family history of colorectal
cancer (no or yes), smoking (never, former, and current),
coffee consumption (no or yes), physical activity (inactive
and active), red meat intake (?5.9 and ?5.9 servings/wk), fi-
34 Nutrition and Cancer 2001

Page 3

ber intake (?16 and ?16 g/day), fruit and vegetable intake
(?40.7 and ?40.7 servings/wk), and total beverage intake
(?2.6 and ?2.6 l/day). All cut points used to create strata
were based on the median value of continuous variables of
the control series. These factors were evaluated because of
their potential to modify the effect of tea consumption on
colorectal cancer risk based on epidemiological or labora-
tory findings.
Results
Table 1 summarizes case-control differences for major
colorectal cancer risk factors. Colon cancer cases were
slightly heavier, consumed less fruits and vegetables and di-
etary fiber, and were more likely to have a first-degree fam-
ily history of colorectal cancer and personal history of colitis
than controls; differences were also noted for education
level, tobacco use, and leisure physical activity. Similar pat-
terns were noted for rectal cancer cases, although they also
consumed less calcium and had more years of chlorinated
water use than controls.
Tea consumption was not uncommon in this population,
with 43% of the men and 51% of the women in the control
group reporting any tea consumption (hot or iced). Equal
numbers of men and women used iced tea (36%), whereas
hot tea consumption was more common among women
(32%) than men (21%). Among controls who drank tea, the
median consumption of any tea was 2.0 cups/day. The re-
spective medians for hot and iced teas were 1.0 and 1.3 cups/
day, with no difference between men and women for hot tea
but a higher median consumption of iced tea for women (2
vs. 1.1 cups/day). There was a weak correlation (Pearson
correlation coefficient = 0.18) between consumption of hot
and iced teas in the control group. Table 2 summarizes the
relation between selected colorectal cancer risk factors and
tea consumption in the control population. Greater tea con-
sumption was positively associated with vegetable and fruit
intake, fiber intake, total beverage intake, and a higher level
of leisure physical activity; there was little evidence for sys-
tematic differences across the other risk factors included in
Table 2.
After adjustment for age and sex, there was no evidence
for a dose-response relation between any type of tea con-
Vol. 41, Nos. 1&2 35
Table 1. Relation of Selected Colorectal Cancer Risk Factors With Case-Control Status, Iowa, 1986–89
Colon Cancer Cases Rectal Cancer Cases
Risk FactorControlsValue
P
Value
P
Mean ± SD
Age, yr
BMIaat age 40 yr, kg/m2
Dietary intake
Vegetables and fruits, servings/wk
Red meat, servings/wk
Calcium, mg/day
Fiber, g/day
Years of chlorinated water
Total beverage l/day
68.4 ±
24.3 ±
9.9
3.5
69.0 ±
24.8 ±
9.3
4.1
0.1
0.01
67.3 ± 10.0
24.9 ±
0.01
0.002 3.8
44.0 ± 22.7
6.9 ±
824± 473
17.1 ±
19.8 ± 18.3
2.8 ±
42.1 ± 23.1
6.9 ±
800± 546
15.7 ±
21.1 ± 18.7
2.8 ±
0.09
0.9
0.3
0.0003
0.3
0.7
43.2 ± 19.6
7.0 ±
775± 428
16.4 ±
22.7 ± 20.5
2.8 ± 1.5
0.4
0.7
0.02
0.06
0.05
0.8
5.84.84.5
7.57.4 6.8
1.21.6
Percent distribution
Family history of colorectal cancer (1st degree)
History of colitis
Education
?High school
High school
?High school
Tobacco use
Never
Former
Current
Leisure physical activity
Inactive
Moderately active
Highly active
Coffee consumption
None
1.0–1.9 cups/day
2.0–3.3 cups/day
3.4–6.0 cups/day
?6.0 cups/day
8.5
11
21
22
0.001
0.001
13
30
0.001
0.001
21
51
28
22
56
22
0.03 24
52
24
0.2
45
39
16
47
42
11
0.01 45
42
13
0.2
50
30
20
55
23
22
0.00755
25
20
0.03
13
21
20
23
23
13
26
19
22
20
0.1 10
24
21
21
24
0.2
a: Body mass index.

Page 4

sumption (hot, iced, and total) and risk of cancers of the co-
lon (all, proximal, or distal) or rectum (Table 3). Adjustment
for potential confounding factors in these data (age, sex, ed-
ucation, leisure physical activity, smoking history, and in-
take of dietary fiber and fruits and vegetables; see Methods)
did not materially change the results (Table 3). At the high-
est levels of tea consumption, there was a weak positive as-
sociation with hot tea (OR = 1.4, 95% CI = 0.8–2.5) and an
inverse association with iced tea (OR = 0.5, 95% CI =
0.2–1.1) and total tea (OR = 0.7, 95% CI = 0.4–1.3), but
none of the point estimates was statistically significant. Fur-
ther adjustment of the associations in Table 3 for body mass
index, family history of colorectal cancer, history of colitis,
total beverage consumption, years of chlorinated water con-
sumption, and intake of red meat, calcium, and coffee also
did not alter the results (data not shown). Adjustment for
season of return of the questionnaire (which could influence
reporting of tea consumption) also did not alter the results
(data not shown).
We next evaluated the interaction between selected colo-
rectal cancer risk factors and tea consumption on the risk of
colon and rectal cancers. There was no evidence for an
interaction with sex, first-degree family history of colorectal
cancer, physical activity, or red meat, fiber, fruit and vegeta-
ble, coffee, or total beverage consumption (data not shown).
However, there was a suggestive interaction with smoking
status for colon, but not rectal, cancer. Among never smok-
ers, there was a suggestive inverse association with tea con-
sumption at the highest level of use, no association of tea
consumption among former smokers, and a positive associa-
tion among current smokers (Table 4). However, the CIs for
all point estimates included 1.0, and the trend tests were not
statistically significant except for the age/sex model for cur-
rent smokers.
Discussion
We found little evidence for an association, either posi-
tive or negative, between tea consumption and risk of colon
or rectal cancer in this Midwestern US population. This was
true for hot tea and iced tea consumption, among men and
36Nutrition and Cancer 2001
Table 2. Relation of Selected Risk Factors With Level of Tea Consumption Among Controls, Iowa, 1986–89
Total Tea Consumption,acups/day
Factor
None
(n = 1,297)
?3.1
(n = 840)
3.1–5.0
(n = 148)
?5.0
(n = 108)
P
Mean ± SD
Age, yr
BMI at age 40 yr, kg/m2
Dietary intake
Vegetables and fruits, servings/wk
Red meat, servings/wk
Calcium, mg/day
Fiber, g/day
Years of chlorinated water
Total beverage, l/day
68.6 ±
24.2 ±
9.9
3.5
68.6 ±
24.4 ±
9.8
3.5
66.9 ± 10.6
24.8 ±
66.5 ± 10.0
25.2 ±
0.04
0.14.0 3.5
42.1 ± 23.1
7.0 ±
826 ± 457
16.8 ±
20.6 ± 18.6
2.6 ±
45.4 ± 21.1
6.6 ±
814 ± 496
17.2 ±
19.8 ± 18.4
2.8 ±
48.4 ± 25.6
7.3 ±
877 ± 460
18.1 ±
16.1 ± 17.4
3.4 ±
50.2 ± 22.9
6.9 ±
824± 493
19.4 ±
16.1 ± 17.2
4.5 ±
0.0001
0.5
0.6
0.006
0.2
0.0001
5.8 5.7 7.44.4
7.76.8 8.09.1
1.1 1.0 1.21.7
Percent distribution
Family history of colorectal cancer (1st degree)
History of colitis
Education
?High school
High school
?High school
Tobacco use
Never
Former
Current
Leisure physical activity
Inactive
Moderately active
Highly active
Coffee consumption
None
1.0–1.9 cups/day
2.0–3.3 cups/day
3.4–6.0 cups/day
?6.0 cups/day
8 10
11
8
9
7 0.6
0.9 11 12
24
52
24
18
50
32
18
51
31
20
56
24
0.001
42
40
18
50
38
12
52
30
18
42
42
16
0.002
53
27
20
47
33
20
42
40
18
41
29
30
0.001
13
18
20
22
27
11
25
23
25
16
15
25
18
23
19
22
24
13
17
24
0.001
a: Cut points for tea consumers based on 75th and 90th percentile distribution of consumption among controls.

Page 5

Vol. 41, Nos. 1&237
Table 3. ORs and 95% CIs for Cancer of the Colon (All, Proximal, Distal) and Rectum According to Level of Tea Consumption, Iowa, 1986–89a
Tea
Consumption,b
cups/day
All Colon Cancer
Proximal Colon Cancer
Distal Colon Cancer
Rectal Cancer
Controls
Cases
OR1
OR2
95% CI
Cases
OR1
OR2
95% CI
Cases
OR1
OR2
95% CI
Cases
OR1
OR3
95% CI
Hot tea
None
1,804
462
1
1
Ref
177
1
1
Ref
272
1
1
Ref
466
1
1
Ref
?1.9
407
136
1.2
1.1
0.9–1.4
55
1.2
1.1
0.8–1.7
77
1.2
1.1
0.8–1.7
101
0.9
0.9
0.7–1.2
1.9–3.0
126
32
0.9
0.9
0.6–1.4
12
0.8
0.6
0.3–1.3
19
0.8
0.6
0.3–1.3
45
1.4
1.3
0.9–1.9
?3.0
56
20
1.2
1.4
0.8–2.5
9
1.4
1.6
0.7–3.4
11
1.4
1.6
0.7–3.4
17
1.1
1.1
0.6–1.9
Iced tea
None
1,537
405
1
1
Ref
165
1
1
Ref
234
1
1
Ref
410
1
1
Ref
?3.0
693
209
1.1
1.1
0.9–1.3
80
1.1
1.0
0.7–1.4
119
1.1
1.0
0.7–1.4
178
0.9
1.0
0.8–1.2
3.0–5.0
99
28
1.1
1.1
0.7–1.8
7
0.7
1.0
0.4–2.1
20
0.7
1.0
0.4–2.1
23
0.8
1.0
0.6–1.6
?5.0
64
8
0.5
0.5
0.2–1.1
1
6
0.2
0.2
0.03–1.5
18
1.0
1.0
0.5–1.8
Total tea
None
1,297
321
1
1
Ref
130
1
1
Ref
185
1
1
Ref
335
1
1
Ref
?3.1
840
261
1.2
1.1
0.9–1.4
101
1.1
1.0
0.8–1.4
151
1.1
1.0
0.8–1.4
214
1.0
0.9
0.8–1.2
3.1–5.0
148
48
1.2
1.3
0.9–1.9
13
0.8
0.9
0.5–1.8
33
0.8
0.9
0.5–1.8
52
1.3
1.4
1.0–2.1
?5.0
108
20
0.7
0.7
0.4–1.3
9
0.8
0.9
0.9–1.9
10
0.8
0.9
0.4–1.9
28
1.0
1.0
0.6–1.5
a: OR1 adjusted for age and sex, OR2 adjusted for age, sex, education, smoking history (never, former, current ?20 cigarettes/day, current ?20 cigarettes/day), leisure physical activity, and intake of dietary fiber
and fruits and vegetables with accompanying 95% confidence interval (CI); OR3 adjusted for the above factors plus coffee consumption with accompanying 95% CI. OR, odds ratio; Ref, reference.
b: Cut points for tea consumption based on 75th and 90th percentile distribution of consumption among controls.